Why Nobel Prize in Chemistry 2013 goes to computational modelers
The natural sciences discovered the power of intelligent machinery, when personal computers were still in their infancy: In the 70s, Martin Karplus, Michael Levitt and Arieh Warshel broke ground for programs that are used to understand and predict chemical processes and received the 2013 Nobel Prize in Chemistry. As quantum chemical methods provide almost chemical accuracy, computational methods are developed to reproduce and rationalize effects and to compare with experimental data. Ansgar Schäfer is responsible for Quantum Chemistry at BASF and explains why the simulation of chemical process is of great industrial importance.
What makes computational modeling so difficult?
Molecules are composed of atoms which are connected by chemical bonds. These bonds are formed by electrons which arrange themselves in molecular orbitals. A proper description of how this works is only possible with the concepts of quantum mechanics. As long as the electron arrangement does not change (in "stable" molecules) the description can be simplified using models from classical mechanics, e.g. springs representing the bonds. These so-called molecular mechanics or force field methods allow a very efficient calculation of large chemical structures like bio molecules. In the field of computational chemistry and biology our colleagues use these techniques to understand and predict the structure and binding properties of active sites in enzymes. Thus new active ingredients for agriculture can be developed.
What does it mean to simulate chemical reactions?
A chemical reaction involves a redistribution of electrons within molecules, in other works breaking and formation of bonds, or between molecules - oxidation and reduction. Here we have to use quantum mechanics to understand and predict such processes. By explicitly calculating all possible structural pathways of how the atoms rearrange in a reaction, we get a detailed microscopic picture of the reaction mechanism. In addition, we can give numbers for thermodynamic and kinetic reaction parameters.
Is computing power an issue for you – in short the bigger the better the computer?
Computing power is primarily an issue for quantum mechanical calculations on larger molecules. The complicated equations describing the behavior of the electrons have to be solved simultaneously for all electrons in the molecule. And the number of electrons increases quickly with molecular size, e.g. the comparably small glucose molecule C6H12O6 contains already 96 electrons. The movement of the electrons is a many-body problem, like the movement of the planets around the sun, which cannot be solved exactly for more than two bodies or electrons. Nowadays we have well-tested approximate methods in very efficient software packages available, which allow us to calculate the structure of a 100-atom molecule within a day and to explore a reaction mechanism in time frames of days to weeks on modern Linux computer clusters. Already in 1999, two other theoreticians, John Pople and Walter Kohn, had received the Nobel prize for their contributions to this development. However, the treatment of really large systems like proteins, which can contain many thousand atoms, is still impossible on the quantum level alone.
Looking at this hurdle, what is the biggest contribution made by the Laureates from your point of view?
The three Nobel laureates have developed a concept to overcome this limitation by combining a quantum treatment of a small part of the molecule with a molecular mechanics description of the big rest, commonly abbreviated as QM/MM approach. This opened the way to study reaction mechanisms e.g. in enzymes, which later on contributed significantly to the knowledge about biochemical processes.
In the layman’s eyes computational modeling means that the computer will replace the classic lab experiment. Is it indeed all about efficiency increase?
From today’s point of view definitely not; there are two important contributions of this type of computer modeling to a research project: First, the calculations give insight which is largely complementary to what can be learned from experiments. The combination of experiment and theory generates a new quality of understanding. Second, a large number of chemical structures can be tested on the computer, even before they exist in reality, e.g. hypothetical new catalysts. This then helps to select the most promising candidates for the experiments. The huge chemical space is narrowed down to an experimentally tractable size. The number of conducted experiments will in many cases probably not change much, but the calculations can guide their selection towards improved probability of success.
The Nobel committee underlined that computer models are used to study industrial chemical processes. Can you give an example from your daily work?
Quantum Chemistry supports the development of improved catalysts, investigates reaction mechanisms, provides data about the thermodynamics and kinetics of reactions, helps in selecting materials with the right properties for organic light emitting diodes or OLEDs and other organic electronics devices, optimizes solvent mixtures and reaction media, guides the selection of porous media like zeolites or metal organic frameworks (MOFs) for separation processes or gas storage, contributes to the development of improved battery materials, and others. But calculation does not stop here - the next step is scientific computing. It closes the gap between understanding of processes on a molecular level to the description and optimization of macroscopic devices, reactors and even chemical plants.
Have you had the chance to work with the Nobel laureates yourself?
Roughly ten years ago, we have been partner in an EU-funded project which had the goal of making the QM/MM simulation techniques applicable to industrial problems of biocatalysis, heterogeneous catalysis and zeolites. Together with Prof. Walter Thiel at the MPI in Mülheim, who is also an important method developer in this field, we have focused on the treatment of reactions in enzymes. The enzyme we have worked on was also investigated by Martin Karplus and his group. We had discussions with him about the methods and the results obtained, and finally published two papers back to back in the same journal (J.Phys.Chem.B 106 (2002) 1758 and 1768).
edited by Anja Feldmann